Waste of resources to ship your own OS?


#1

Why does PureOS exist?

The way I see it is that your added value is in providing hardware and platform support (a Linux kernel concept).

I have absolutely no interest for example in running a Debian derived system, because of the work involved in keeping my system up to date. In particular upgrades between major versions generally do not work.

I fully expect that if the Purism laptops ever become popular that a similar thing could happen. Offering a supported OS is a huge support cost and I think you shouldn’t offer that service.

Pop!_OS from System76 is a similar mistake. Even if it would work perfectly (which it can’t), it would still be a slightly different operating system from our production environment. I can imagine that many developers would opt for a native operating system closest to whatever is running on their servers.

I can sort of see the need to ship something, but a rebranding seems to be the wrong strategy.

I would like to see you commit to making all your patches available for easy, automated integration into other Linux (and possibly others) distributions with all the platform support in the Linux kernel upstream.

When I look in the Linux kernel and search for librem or purism, I don’t see anything returned, which seems to suggest you do not have special support available, like there is for example when you compare to one of the best supported devices for Linux that has ever existed (the thinkpad).

See these links to verify:
https://livegrep.com/search/linux?q=librem&fold_case=auto&regex=false&context=true
https://livegrep.com/search/linux?q=purism&fold_case=auto&regex=false&context=true
VS
https://livegrep.com/search/linux?q=thinkpad&fold_case=auto&regex=false&context=true

The laptops I use are more expensive than the ones you are offering, so it’s not a money issue. You are doing a lot of things right, but I also wonder a bit about your lack of focus in that you are also trying to develop a phone now. You have 14 employees, not 150.

I hope you appreciate my feedback and focus your development efforts on making differentiating hardware features while supplying generic support for Linux and get rid of PureOS as much as you can.


Alternative to Flathub
Librem 5: Desktop Decision and Development Efforts
#2

a similar aproach is done by qubes, the packages added are an ugly hack that does not build for others and are useless outside the qubes build. this is very sad and hinders freedom of choice (if not being a violation of gpl).

i wonder why companies go that way instead of only providing small, reusable units that can just be used in many environments…


#3

I would like to hear more about how Qubes does things and why you think their behavior violates the GPL.

There is always the option that it is incompetence and not malice. It would certainly be interesting to see an official statement.


#4

qubes is a bit off-topic here but maybe it applies to purism as well.

about GPL: the GPL states that a package must be able to be build which is not the case for the individual package. the sources include distribution dependend build instructions mixed into the whole build process (which as far as i know is not much loved by debian, i don’t know about others).

i doubt we can talk about incompetence when we talk about qubes or purism because these people do an incredible good job. i assume marketing and resource shortage are cause for this. the end user expects a smooth expirience and loves eye candy with a PureOS logo on his new pure notebook. this can be achieved fast & easy by an own distro (looking at ubuntu, the big paragon). so myself as a technician tries different stuff and feels the pain.

point is: you can fund such projects if you get enough people in and we technical nerds are just not enough :wink:


#5

I agree. Instead of yet another distribution, the team should imho focus on the HW, firmware and low-level programming (that’s a big deal, costly, and hard task!). You could leave the most of the higher level programming on the community.
Ideally just orchestrate actions between Debian/Ubuntu, KDE/Gnome teams etc.


#6

including some resources on documentation specifically for Librems with some std distros so that enthusiastic noobs like myself are able to join in the fun without having to do a crash course in linux to get up to speed on a Librem.


#7

Guys… please… take the time and find out what Purism is actually about, maybe then you can understand what they do and hopefully appreciate it.

Yes, there are way too many distros and Pop!_OS is probably just a useless Mee-Too distro.
But PureOS is not about just rebranding something.

Purism was founded specificially to make it easy, even for non-technical people, to run 100% free, privacy protecting systems. This statement alone should make it clear that Debian is not the way to go. If it does not, have a look at the Purism freedom roadmap.
It might surprise you but Debian is actually not endorsed by the FSF. Here’s a list of endorsed distros, which is surprisingly short.
Purism strives for FSF endorsement for all their hardware and software, and therefore for PureOS 3.
Now, with the upcoming Librem 5, PureOS will also become a mobile OS, with built-in convergence.

Purism has an upstream-first philosophy, so all general improvements will sooner or later be part of other distros. But some modifications like a privacy-by-default browser might be to specific to be adopted by others. You cannot - and should not - impose your philosophy on others. Therefore, I think they made the right choice.

Also, that freedom Purism is striving for includes the freedom for you to replace the default operating system with whatever you like.


#8

Actually I agree with the decision that both Purism and System76 have made regarding forking and re-badging a distro the work that both are doing would be the same regardless of weather they gave the distro a unique name. what the fork enables both companys to do is have the customized patches there that are unique to the respective company hardware without waiting for the patches to be accepted up stream. yes the patches could be in the OEM Image, however, i do feel that both have done things correctly.


#9

Purism was founded specificially to make it easy, even for non-technical people, to run 100% free, privacy protecting systems.

That’s the problem. There are two different sorts of customers: technical people and non-technical people. Currently, the customers attracted to Purism are technical people who are looking for a laptop with the Intel Management Engine disabled, for example. But Purism wants to attract non-technical people who don’t even know what an “Intel Management Engine” is. Presumably, this is because there are relatively few technical people in the world, and Purism could potentially make a lot more money if it appealed to non-techincal people.

This leads to a conflict of interests: the current customer base (techinical people) want one thing, but Purism wants something else to appeal to the non-technical people.

The fact that Purism is developing PureOS and the Librem-5 phone clearly indicates that they are trying to appeal to non-technical people. The fact that the Purism laptops don’t have built-in ethernet ports, for example, is an indication that purely techincal people are not the desired customers.


#10

Sorry, but no, that’s not the problem. It’s their mission. Purism of course knows that currently their main audience is techy people. Like themselves. I’d say non-tech people are a long-term goal, but not the current main focus. It’s funny how people take the same facts to prove opposing points. Others suggested that the Librem 5 will fail because it targets a nerdy niche and not mainstream.

Of course it’s sad that there are no Ethernet ports on (most) Librems. But you cannot just take a random fact to prove a random theory. How about they are small, hardware design is complicated, there are so many wishes customers have and it’s hard to make everybody happy? And, I have to say… Qubes users that think Ethernet-over-USB is potentially less safe than plain Ethernet unless that USB is separated from all other USB devices, that is really a special kind of tinfoil users. A niche of the niche. If they target you specifically, they’re bankrupt before I get my Librem 5 :frowning:
Give them time to grow and maybe one day the Ethernet port will be back in some of the models.

We got very off-topic here.


#11

This is true if you make the assumption that Purism primary goal is to make money on the back of rising awareness of the issues related to security and freedom of the Internet.

However, it is also possible that this ain’t their primary goal, whereas their aim is first and foremost to create a a quality product that respects your online freedom, and therefore charging for the cost involved in order to make it economically sustainable.

Seeing the extent of their roadmap would make me think that the later is the most likely scenario, for I think that even with the current coreboot implementation they could turn their attention to polishing up the product to attract larger number of clients and rake-in the moolah if their goal was the former.


#12

They could have picked GuiXSD.

Just because you have some extra packages, doesn’t mean you have to do all the branding differently.


#13

Do we all just conveniently forget that Purism is a social purpose corporation?

Profits are a thing, sure, and you’ve got to have them for the company to work unless you’re getting donations from somewhere. People don’t do things for nothing, and you can’t buy parts or do research without it either.

However, I don’t think that’s their primary motivator. If it is, it’s not supposed to be.

Anyway, I don’t really take Purism as to be some greedy megacorp. I do hope their prices are lowered soon, but I assume there are reasons for it beyond profit margins.


#14

Yes.
But why? Debian is probably updated more actively and that’s probably the reason why so many distros are based on Debian. Also, most Purism developers are experienced Debian developers, so it’s a natural (and efficient) choice.

Just because you have some extra packages, doesn’t mean you have to do all the branding differently.

“What OS do you have on your box?”
“Debian!”
“Vanilla Debian or Purism Debian?” <- - - That question you’ll never hear
“Purism Debian of course. I want the privacy extensions, but not the binary blobs”
“How do I install it?”
“Oh, simple. Install vanilla Debian, remove some extensions, add Purism repo, install 50 Purism packages. Shouldn’t take you more than an hour if you’ve done it a few times. One time I got a dependency conflict but usually it’s fine”
“Uh… why didn’t they just make a derivative like everybody else?”
“I think they wanted to save the hassle of exchanging some logos”
“Seems legit”


#15

Yeah, perhaps I am overestimating the amount of branding that needs to happen. Is Iceweasel (as an example) trademarked?

Having said all this, I would expect platform support (like the ThinkPad) in the Linux kernel for Purism.


#16

It’s funny that you bring up Iceweasel, which has been just a rebranded bundle due to Trademark issues. According to Wikipedia, that dispute is over. Iceweasel is explicitly not trademarked, and Mozilla probably relaxed some requirements when using their trademarks.

Having said all this, I would expect platform support (like the ThinkPad) in the Linux kernel for Purism.

Purism goes the other way: As far as possible, use components that are already (well) supported by the kernel. If not, of course it’s added upstream.


#17

I definitely agree with the post that started this whole thread. When I was considering whether I should buy it or not, I was kind of thinking “how much of my money will go into funding distro (n+1)”? I still jumped in a bought a Purism laptop, but this was definitely a minus point for me. And for sure, I never even considered using PureOS. I use my laptop for work, I want the OS to be similar (or even a release ahead) of what I use most of the time on my other systems. And I fear like hell “niche” infrastructure software (I’ve got burned a few times with infrastructure software that turned out not to be supported by enough people to be viable on the long term). On the other hand, I understand the point that Purism wants to be able to offer a secure and smooth experience, but I don’t believe for a minute that non-tech users would buy Purism products (non-tech users go to best-buy to get a laptop “like the one they saw on tv”).

Now, I wish that Purism were much more proactive in pushing patches upstream. The touchpad of my Librem15v2 is still not properly supported with the upstream driver (compared to its driver in PureOS) and I have pretty much lost all hopes that this would change one day.


#18

Can you make the same argument with easy to check references (like I did in my post)? Your post is demonstrating exactly the kind of thing I want to avoid funding.

Your post is also a direct contradiction of what @Caliga said.

If both can produce evidence, we might be able to label one of the two as dishonest.

Anyway, I think that because we think the same about this topic, I strongly suspect that there actually is a group with this mindset, because it makes economic sense. It is cheaper for the users to be able to select any Linux distro and so there is more value.

Purism could simply state that any patch will be offered upstream within two months and reasonable efforts to make the code work to the standards expected by the Linux kernel project.

We are living increasingly in a world where consumers get to dictate the products they want. We are clearly telling you what we want.

I am fairly influential with regards to buying laptops even in bulk; if Purism would happen to be selling what I am interested in at the time the next bulk contract gets negotiated, I could advocate in favor of Purism and I am sure there are many others for which this would hold. Your business actions do have a material impact on the decisions made by other economic actors.

Things I would like to see before that happens are:

  • a case-study of existing business users (those that travel a lot) showing how happy they are with their devices comparing it to previous devices they have owned
  • open bulk pricing (so, we don’t both have to waste time on negotiation)
  • metrics displaying amount of code that is not yet in Linux upstream, etc. on a status page, metrics showing shipping time, service time, MTBFs, etc.
  • noise level data comparing to other top of the line laptops (quiet laptops are good for productivity)

I am writing this, because I hope you succeed and you do the right things. If not, the next company trying to do these things already has some requirements to work with.


#19

Hi!
You can have a look at an older thread where I mentioned several issues, including that although the touchpad works, it does not has the same features between PureOS and upstream: Review of Librem15 v2.

On PureOS, the touchpad has multi-touch (so two fingers scrolling works), but not on other distributions (I am now using the latest Debian). There are quite a bunch of patches floating around, so somebody put a project on github to collect all these patches: https://github.com/kylerankin/psmouse-byd-dkms. And then, a thread on these forums where the developer of the upstream driver jumped in: BYD touchpad multi-touch support?. You can see the mismatch between the features found on PureOS and the upstream…


#20

@bavay it seems to me that the touchpad issue is a perfect counter example. I did not exactly understand whether the problem is that other distros don’t have BYD enabled by default or a patch was not accepted by upstream or both or whatever. But that is exactly the trouble you have to deal with: Push to upstream, hope they accept it, rework it, submit again, beg them, ask them… Then, when it’s in upstream, ask (or wait for) distros to pick it up… That is time consuming, frustrating and not predictable. You easily end up with a situation like:

Today we announce the availability of the Librem 42. It mostly works with Debian. Our drivers did not make it into Fedora 27, but we expect them to be in Fedora 28 so check back in 6 months.
Or download our modified Fedora-kernel. And yes, of course we will release kernel updates for Fedora 27 whenever the original kernel is updated. Promise!

Not only do you not save any time by this, you can just hardly release a complete product if you rely on external dependencies.
Also, as a reminder:

  • Purism wants a FSF endorsed distro to ship by default (see my first post)
  • Purism only uses hardware without good mainline kernel support if that brings them closer to the blob-free goal
  • The Librem 5, including desktop convergence, are simply not possible with the above model

Pushing to upstream is not the same as having it accepted by upstream and not the same as being adopted by other distros. Still, if you see a lack of pushing to upstream, then friendly remind Purism that they have an upstream-first philosophy. I think that’s more important than giving up PureOS.

Also, I still think you are greatly overestimating how much resources are needed to take an existing distribution and enrich it with self-developed enhancements (which are part of the security, privacy and freedom philosophy Purism follows). The (outdated) team page lists 3 PureOS developers. Let’s assume one of them does nothing else but maintain PureOS and the other two do the actual development. That makes the cost of maintaining the OS (instead of just developing packages) seven percent of staff cost.
If you assume that the pure hardware cost is 50%, then you pay 3.5% for PureOS to exist. That is $56 for the Librem 15. I think that is a worst-case scenario. $15…$30 seems more realistic to me. But let’s say $56.

Now, you double the order quantity. The effect is very likely that hardware costs are reduced by $100. I can just guess here, but the first Librem was $1899 instead of $1599. Also, of course the PureOS share is cut in half, too. So then it became $28 in my worst case.
Anyway, it should be clear that more orders are a significantly better lever to reduce the cost than giving up PureOS (which I still think is not even a viable idea).

@cinderella, therefore I kind of agree with you that bulk orders are something that needs attention. I would offer bulks up to 10 or 20 items. “Contact us for higher quantities”
If it’s more, I guess most people would want to negotiate anyway, for example off-shore or not plays a role, or if you order 1000, do you need them all at once? I’m almost certain you’ll even get whatever keyboard layout you want if you take 1000 :wink:
Also, for the really big numbers it might be smarter to schedule them for the next batch instead of taking them from stock: By adding 1000 to the next batch, that whole batch would become a lot cheaper, so the discount can be higher.

If you are really serious about a bulk order, do them a favour and directly ask for a quotation, if it’s more than 100 ask if it makes sense to wait for the next batch and request the metrics and stuff you mentioned. They can only afford to invest that time if there will be some outcome, which is very understandable given the size of the company. That way you help them much more than by telling them how to run a business.

Just my two cents :wink:

[EDIT]: Most of the topics touched here, from “why PureOS” to developing BYD touchpad driver and submitting it to the kernel (7:00) to original Librem prices, leverage by ordering bigger quantities are explained in detail by @todd-weaver in the video on https://puri.sm/about/